Stochastic Stability of Monotone Dynamical Systems. I. The Irreducible Cooperative Systems
Jifa Jiang · 2024
Read Stochastic Stability of Monotone Dynamical Systems. I. The Irreducible Cooperative Systems by Jifa Jiang (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-19977.txt.
Introduction
Jiang, Sheng, and Wang study how orderly systems behave when noise is added, in this two thousand twenty-four paper. It opens a series on monotone dynamical systems, which model processes where pushing one part up never pushes another down, from chemical networks to populations.
This first part treats cooperative, irreducible systems and proves a strong form of stability: as noise shrinks, the system's fate concentrates on a simply ordered set of steady states where nothing changes. The proof leans on large deviation theory to price the rare paths near tangled invariant sets. Applied to classic positive feedback systems, the result says the long-run outcome lands on a finite menu of stable equilibria, even when cycles exist.
Catalog record
| CREATOR | Jifa Jiang |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-19977
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Stochastic Stability of Monotone Dynamical Systems. I. The Irreducible Cooperative Systems Thanks: Supported by the National Key R&D Program of China (No.2024YFA1013603, 2024YFA1013600), the National Natural Science Foundation of China (No.12331006), and the Strategic Priority Research Program of CAS (No.XDB0900100). Jifa Jiang Abstract The current series of papers is concerned with stochastic stability of monotone dynamical systems by identifying the basic dynamical units that can survive in the presence of noise interference. In the first of the series, for the cooperative and irreducible systems, we will establish the stochastic stability of a dynamical order, that is, the zero-noise limit of stochastic perturbations will be concentrated on a simply ordered set consisting of Lyapunov stable equilibria. In particular, we utilize the Freidlin–Wentzell large deviation theory to gauge the rare probability in the vicinity of unordered chain-transitive invariant set on a nonmonotone manifold. We further apply our theoretic results to the stochastic stability of classical positive feedback systems by showing that the zero-noise limit is a convex combination of the Dirac measures on a finite number of asymptotically stable equilibria although such system may possess nontrivial periodic orbits. Keywords : Stochastic stability; Freidlin–Wentzell’s large deviations principle; Cooperative systems; Zero-noise limit; Lyapunov stable equilibria. 1 Introduction Monotone dynamical systems, stemming from the groundbreaking work of M. W. Hirsch, are characterized by the presence of a comparison principle that is aligned with a closed partial order relation (induced by a convex cone) within the state space. Such systems, due to Smale [ 59 ] , may harbor orbits that exhibit arbitrary complicated behavior, including horseshoes (see also in e.g., [ 11 , 63 , 62 , 67 , 68 , 71 ] and references therein). On the other hand, the signature results of monotone dynamical systems indicate that the forward orbit of almost every (i.e., generic/prevalent) initial state converges in both the topological sense ( [ 49 , 51 , 19 , 21 , 17 , 60 , 66 , 69 , 70 ] ) and measure-theoretic sense ( [ 13 , 72 , 73 ] ). Over the past decades, extensive research has been conducted in this field, and its applications have been continuously expanding. For more detailed information, one may refer to the monographs [ 18 , 61 , 23 , 64 , 50 , 57 , 3 ] as well as their cited literature. Numerous mathematical models, originating from differential and difference equations, can give rise to monotone dynamical systems (see [ 61 , 64 , 23 , 79 ] ). Given that these models are frequently prone to noise-induced perturbations, the generated monotone dynamical systems are inevitably affected by a wide range of irregularly occurring phenomena. As a result, the study of how noise perturbations affect the dynamics of monotone systems has become a fundamental concern, critical to both theoretical modeling and the practical investigation of system behavior. Prior research has addressed this dynamical issue, mainly concentrating on a trajectory-based approach that is predominantly utilized within the framework of random dynamical systems. One may refer to [ 1 , 7 , 8 , 14 ] for recent developments in the theory of monotone (order-preserving) random systems, as well as the application to the long-time behavior of random and stochastic differential equations. For instance, Arnold and Chueshov [ 1 ] studied certain simplification in the long-time dynamics for random monotone systems. Synchronization by noise was subsequently observed in random monotone systems (see Flandoli et al. [ 14 ] , Chueshov and Scheutzow et al. [ 8 ] ), which indicates that there is a certain point attractor consisting of a single random point and in this sense the random dynamics are asymptotically globally stable. Thus, it becomes clear that the long-time behavior of a general monotone random systems is notably simpler when compared to its deterministic counterpart. In other words, the trajectory-based approach might not yield much insights into the noise-induced perturbations within monotone dynamical systems. To our current understanding, in the realm of general noise perturbations, the distribution-based approach stands out as an especially potent strategy. It has become an increasingly essential instrument, providing greater utility than the trajectory-based method. A fundamental problem in the distribution-based approach is to identify the basic statistic dynamical units (for instance, certain invariant measures and their concentrations) which are stochastically stable , that is, they can “survive” in the presence of noise interference. Following L. S. Young [ 10 , 77 ] , we called such invariant measures as zero-noise limits of random perturbations of dynamical systems , that is, zero-noise limits are the weak limits of \mu^{\varepsilon} as \varepsilon\to 0 . Here, \mu^{\varepsilon} denotes the probability measure that is invariant under the Markov process corresponding to noise level \varepsilon . These ideas go back to Kolmogorov (see [ 58 , p.838] ). The importance of zero-noise limits is clear: assuming that the real world is perpetually slightly noisy, they represent measures that are the truly observable invariant measures. Elucidating the precise locations for concentration of zero-noise limit is essential for the discernment of invariant sets exhibiting stochastic stability. For some pioneer works from this point of view, we refer to [ 39 , 47 ] for stochastic stability of flows on a cycle or a 2-torus, to [ 28 , 29 , 53 , 78 ] for stochastic stability of (quasi)attractors, to [ 42 , 41 ] for stochastic stability of hyperbolic or axiom A systems, and to [ 10 , 76 , 77 ] for SRB measures as zero-noise limits with positive Lyapunov exponents, etc. Moreover, we refer to Freidlin and Wentzell [ 74 , 15 ] for the stochastic stability of systems whose long-time behavior can be expressed as a finite union of certain basic sets, termed equivalence classes (with respect to the quasipotential) in their works. This series of our work will adopt the distribution-based approach to study the impact of white noises on the dynamics of monotone dynamical systems. Since zero-noise limits mirror the stochastically stable behavior, one cannot solely rely on the analysis of the long-time behavior of individual orbits. Accordingly, due to its essential differences from deterministic monotone dynamical systems, much less is known in this direction comparing with cases using the trajectory-based approach. In the present paper, we mainly focus on the zero-noise limits for the stochastic perturbation of a prototypical example of monotone dynamical systems, that is, the cooperative ordinary differential equations (see [ 19 ] ): \frac{\mathrm{d}x}{\mathrm{d}t}=b(x),\quad x(0)=x\in\mathbb{R}^{r}, (1.1) where b:\mathbb{R}^{r}\to\mathbb{R}^{r} is continuously differentiable. Cooperation means that an increase in one component of x(t) causes an increase of all the other components, modeled by the assumption that \partial b_{i}/\partial x_{j}\geq 0 for i\neq j . Moreover, we further assume that the Jacobian matrices Db(x) are irreducible , which means that each component of x(t) directly or indirectly affect all the others. Such systems occur in many biological, chemical, physical and economic models. For the stochastic perturbation of system ( 1.1 ), we consider the stochastic equations \mathrm{d}X^{\varepsilon}_{t}=b(X^{\varepsilon}_{t})\mathrm{d}t+\varepsilon\sigma(X^{\varepsilon}_{t})\mathrm{d}W_{t},\quad X^{\varepsilon}_{0}=x\in\mathbb{R}^{r}, (1.2) where the perturbation parameter \varepsilon is small, and W_{t}=(W^{1}_{t},\cdots,W^{r}_{t})^{T} is a standard r -dimensional Wiener process. The diffusion matrix \sigma is locally Lipschitz continuous and non-degenerate in the sense that a(x)\triangleq\sigma(x)\sigma^{T}(x) is positive definite for